Heterogenous Network Backup

Over the time I managed to get me some computers:

  • 1 Linux Server with 1 Virtual Guest (both debian linux)
  • 1 Desktop Computer with a multiboot achitecture (OSX Lion, Windows 7, Linux Mint 12)
  • 1 Macbook with OS X Lion
  • 1 Server in a remote data center with 4 virtual guests (xen, all 5 systems debian linux)

For the OS X based computers I set up a netatalk (AFP) share so they could backup up via their integrated time machine feature. For the root servers I used a self baked bash script that wrote the important data onto an ftp server also staged in the data center. The data stored on my home server I also wrote manually from time to time onto an external drive. WHAT A MESS.

At my job we are using amanda for backing up a couple of linux servers. It always made me sick thinking about the complexity due to the fact that those programm managed it’s data over virtual tapes even though we used hard disk storage. It was quite a challenge just to figure out:

  • how many virtual tapes do we need?
  • what size should the tapes be?
  • how often to do a full backup?
  • and how many data this will probably consume on the backup server

In addition to that amanda decides autonomously when  actually to do a full backup in order to keep the transfered data and also the time used to a minimum. That might be helpful in some scenarios but in my opinion it’s not very effective when you want to determine how far you can go backup with your restore.

So I began to search for other possibilities that should meet the following requirements:

  • Easy to setup
  • Easy to understand ( IMO that’s absolutely crucial)
  • Easy to restore
  • Reliable (of course)

I came across Bakuppc which blew me away since I’ve been using it (like 2 weeks). Backuppc does not have these relic virtual tape crap. It just saves to disk which is what I want. What I like most about Backuppc is the following facts:

  • It’s very flexible since it relies on very common system features like tar, rsync and ssh
  • It has a nice Web Interface with which I can handle every day business quite easily + it gives me nice statistics
  • It finds identical files through all my different backup sets and keeps the used disk space to a minimum by making hard links

What I don’t like:

  • It’s not able to backup my IPv6-only host since backuppc (3.2.1) does not find an A-Record (even though there is an AAAA) Ubuntu Bug
  • Backuppc has quite good features to schedule, but it has not the ability to give a specific day or time when to backup

What I did not do until now is a restore. That’s a very crucial point which is in many cases not being tested. And when it comes to a failure you might look very stupid. Beyond that I’ll try to make backups of an openwrt Router (Backfire) and a jailbroken Apple TV 2. I try to keep you updated.

Here you can see two screenshots of backuppc’s web gui:

Server Status:

 

 

 

 

 

 

 

Host summary:

 

 

 

 

 

 

After using this peace of software for some month now, I can conclude:

– annoying reinstallation of software after server crash in order to restore
+ very flexible because ssh and smb backup seem to be sufficient for every case
+ good to handle web user interface
– sometimes struggling with file permissions or manual sudo corrections
| better be used in smaller environment (like upto 20 PCs.)
| handy deduplication if there’s enough server power
+ proves that it IS possible to get a much better retention time with less than the expected amount of memory